Why you shouldn’t trade based on political polls

Politico is out with a new poll and it once again show a very tight race that comes down to Pennsylvania, which is near deadlocked.

Pennsylvania: Harris 48%; Trump 47%

Michigan: Harris 49%; Trump 45%

North Carolina: Trump 47%; Harris 45%

New Hampshire: Harris 50%; Trump 43%

Yesterday, we saw bonds sell off and the US dollar rally late in the day after a Quinnipac poll showed Trump making progress in that state. The poll was particularly notable because the same pollster had previously shown him trailing.

Some people in financial markets appear to be making bets on the election but the wiser ones are waiting for the dust to settle. Vanderbilt University political scientist Josh Clinton illustrates how polling in the US is difficult. Pollsters underestimated Trump before and have tried to improve their methods but it ultimately comes down to a series of assumptions.

” Simple and defensible decisions by pollsters can drastically change the reported margin between Harris and Trump,” he writes, showing that national margins could change from Harris +0.9% to +9% based on those.

“One way to address this is by using voters in past elections as a
benchmark. But which election? The 2022 midterm election was most
recent, but midterm voters differ from presidential voters. The 2020
election could be a better choice, but perhaps the pandemic made it
atypical. Maybe 2016 is better.”

You get the idea and when you layer on things like likely voters, hidden party association and asking how people voted last time, you end up with a milt-variate picture that’s far wider than the simple numbers from polls indicate.

What’s the takeaway?

“We would all do better to temper our expectations about preelection
polls. It’s impossible to ensure that the polls will reliably predict
close races given the number of decisions that pollsters have to make.
And it’s often hard for consumers of polls to know how much the results
reflect the opinions of the voters or the pollsters.”

Don’t be surprised if there’s a big polling miss once again on election night, in one direction or the other. Let’s at least hope that makes for a clear result.

This article was written by Adam Button at www.forexlive.com.